The supply chain has been significantly impacted recently, and as a result the handling of products at the delivery and supply stage is of great importance. Currently, the agronomic industry is one of the most studied, but the reliability of harvests is not usually evaluated. This study uses a novel reliability analysis with an accelerated life test to expose the European cucumber (Cucumis sativus L.) to accelerated temperature and humidity conditions. The objective is to observe the effect of both factors on the deterioration of the product. The analysis includes a degradation analysis to determine the significant factors causing degradation, followed by an accelerated life test (ALT) to determine the product’s shelf life. Finally, through the development of a reliability model, storage times for cucumbers under normal storage conditions are predicted.

This investigation presents a data-driven Long-short Term Memory battery model for predicting State of Charge for lithium-ion batteries LiFePO4 for next-generation vehicle operations. Our modified algorithm builds and updates a model using multivariate inputs that include physical properties, voltage, current, and ambient temperature during operations. The primary research goal is to improve prediction performance on future values from multiple training examples using an online learning scheme. Initial results demonstrate excellent predictions that outperform results from literature and other neural network algorithms. Due to computing constraints in on-board vehicle systems, the authors develop online training with autonomous control of lag (window width). The control algorithm embeds in the model with rules that govern and adjust lag during training. This method ensures the minimization of computational cost and prediction errors with the use of standard computing equipment during driving conditions.

Origin-bound password derivation makes a phished credential useless anywhere except the site on which it was captured, and it can be deployed with no server-side changes. The most complete prior client-only realization, AnchorPass (Al-jbour, 2026), nevertheless carries three practical costs that block adoption: a device-held secret (pepper) whose loss silently invalidates every account with no recovery path; a memory-hard hash recomputed on every login; and full-origin binding that breaks legitimate multi-subdomain sites. We present Harbor, a Chromium-based browser prototype whose trusted chrome contains a native credential bar below the address bar, and ROBD (Recoverable Origin-Bound Derivation), a scheme that resolves all three costs without weakening the security guarantee. ROBD replaces the pepper with a 128-bit recovery seed rendered once as a checksummed recovery code, preserving the same 2^{H\left(P\right)+128} offline hardness while making every site password recoverable on any device. It amortizes memory-hardness by running Argon2id once per unlock to produce a root key, then deriving per-site tokens with HKDF in under a millisecond. It scopes tokens to the registrable domain (eTLD+1) with public-suffix–aware separation of hosted-subdomain platforms, and separates HTTP from HTTPS. Because the credential bar lives in browser chrome, the master password is never typed into a page, and a credential firewall blocks the case where a user does so anyway. Across measured experiments—1,000 domain-binding trials, 1,005 cross-domain replay and adversary-in-the-middle (AiTM) relay attempts spanning 15 phishing techniques, latency benchmarks, an offline-cost analysis, a 12-pattern login-UI compatibility suite, and a 200-trial recovery test—Harbor achieves 0.0% replay success, 0.05 ms mean per-login derivation (vs. approximately 118 ms for the prior work), exact recovery in 200/200 trials, and correct field handling on 12/12 UI patterns. Harbor attains passkey-level phishing resistance while retaining password-level deployability and eliminating the loss and performance penalties of prior client-only schemes.

The traveling salesman problem (TSP) is the canonical combinatorial optimization problem famous throughout literature. There exists an objective function associated with every feasible solution. However, the increase in the number of possible solutions makes this an NP-Hard problem. We show that the central limit theorem (CLT) applies to the problem. We then conduct extensive computational testing to show that the cycle lengths tend to a normal distribution as the problem grows large. When the size of the TSP problem exceeds computational power, better understanding solution distributions allows us to save resources. This is a non-trivial result as understanding solution distributions in huge TSP problems helps us to minimize computational effort that may not lead to significantly better results. El problema del agente viajero (“Traveling Salesman Problem” o TSP) es el problema canónico de optimización combinatoria usando en la literatura. Existe una función objetivo asociada con cada solución factible. Sin embargo, la tasa de aumento en el número de soluciones posibles hace que este sea un problema NP Difícil. Mostramos que el teorema del límite central (Central Limit Theorem o CLT) se aplica al problema y luego realizamos pruebas computacionales extensas para mostrar que las longitudes de los ciclos tienden a una distribución normal a medida que el tamaño del problema crece. Cuando el tamaño del problema TSP excede el poder computacional, una mejor comprensión de las distribuciones de soluciones nos permite ahorrar recursos. Este es un resultado no trivial, ya que comprender las distribuciones de soluciones en problemas TSP enormes nos ayuda a minimizar el esfuerzo computacional que puede no conducir a resultados significativamente mejores.

The rapid advancement of mobile devices has significantly increased power consumption, presenting major challenges to energy efficiency and sustainability. As mobile devices become central to daily life, optimizing power management has emerged as a critical research priority. This paper provides a comprehensive narrative review of Artificial Intelligence (AI) techniques, including Machine Learning (ML), Deep Learning (DL), and Reinforcement Learning (RL), for optimizing power consumption in mobile devices. A narrative review methodology was employed, searching IEEE Xplore, ACM Digital Library, ScienceDirect, and Google Scholar databases using keywords including "AI power optimization," "machine learning battery management," and "mobile device energy efficiency." Studies published between 2020 and 2025 were included, focusing on peer-reviewed articles addressing AI-based power management in mobile computing environments. The analysis reveals that ML approaches demonstrate improvements in battery life prediction accuracy, while DL techniques excel at modeling complex, non-linear power consumption patterns. RL methods show particular promise for real-time adaptive power management, with studies reporting battery life improvements ranging from 12% to 27% compared to traditional approaches. However, significant challenges remain, including data requirements, real-time optimization constraints, hardware limitations, and scalability across diverse device platforms. AI-driven power management represents a promising frontier for mobile energy optimization. Future research directions include the integration of federated learning for privacy-preserving optimization, quantum computing for complex optimization problems, IoT-based coordinated power management, and the development of cross-platform AI models. Additionally, research should focus on developing lightweight models suitable for resource-constrained devices, creating standardized benchmarks for evaluating AI power management solutions, and exploring edge-cloud hybrid architectures to balance computational demands with energy efficiency.